Artificial intelligence in fluidisation technology: promise, progress and pitfalls

Artificial intelligence (AI) has caused, and likely will continue to cause, seismic shifts in the way we live, work and conduct research. The field of fluidisation technology is by no means insulated from this shift. AI represents a tremendous opportunity to better understand and improve a wide variety of industrial processes, but also presents significant risks. If we in the fluidisation field wish to benefit maximally from the AI revolution, it is vital that both opportunities and risks are fully recognised, but that the former are not overstated, nor the latter understated. This brief perspective aims to provide a broad and balanced overview of the types of AI tools which are, or soon could be, implemented in fluidisation systems, outlining the state of the art, assessing maturity and highlighting the most significant issues and limitations. It examines the use of AI to develop data-driven predictive models; augment, accelerate or even entirely replace numerical simulations; enable process monitoring, control and digital twinning; automate post-processing and analysis; and optimise operating conditions, formulations and even equipment geometry. Although pilot- and commercial-scale studies demonstrate tangible value, the literature remains dominated by laboratory-scale work. Wider adoption will require rigorous industrial validation, improved AI literacy and greater attention to environmental and ethical impacts.